Title :
Research on Several Problems in BP Neural Network Application
Author :
Huang Duo ; Zhao Xinyu
Author_Institution :
Sch. of Bus. Adm., Xi´an Univ. of Technol., Xi´an
Abstract :
Four problems existing in BP neural network practical application are discussed in this paper: slow converging speed, converging at local minimum, overtraining and poor extrapolating capacity. Comparing advantages and disadvantage of several acceleration algorithms, the paper provides the suggestions of solving those problems. The paper also advances a forecasting model of monthly irrigation return flow based on BP neural network is advanced as a case study.
Keywords :
backpropagation; extrapolation; forecasting theory; irrigation; neural nets; BP neural network application; acceleration algorithms; extrapolating capacity; forecasting model; monthly irrigation return flow; Acceleration; Arithmetic; Computer errors; Heuristic algorithms; Intelligent control; Irrigation; Neural networks; Pattern recognition; Predictive models; Water resources;
Conference_Titel :
Wireless Communications, Networking and Mobile Computing, 2007. WiCom 2007. International Conference on
Conference_Location :
Shanghai
Print_ISBN :
978-1-4244-1311-9
DOI :
10.1109/WICOM.2007.1647
